Coding/AI Best Practices

Coding standards and style guides

Write for humans, not machines - Code should be readable like a story.
Good code doesn’t need to be commented!
Add documentation comments and types where helpful.
Use semantic namings for variables - think from the dev perspective.
- For example: bot.run() instead of bot.run_bot()
Avoid magic numbers or strings. Extract them as constants. Use enums. Don’t hardcode vars.
Be consistent in formatting and styling your code.
If you doubt your skills - ask LLMs and seniors to check.
If you use LLMs to write code, make sure you understand its output.
Don’t share private information about the project with LLMs.
Python writing style guide to follow (from google): https://google.github.io/styleguide/pyguide.html
When adding a TODO item in code, use the format: TODO: {Issue Type}/{Issue ID}: {Short Description}. If the issue is not in the same repository, add a link or full reference to the repository.

Use it when it helps, but not more.
Turn off data collection and do not use AI that is collecting data from your prompts. Especially for AI-integrated editors.
Do not rely on AI’s expertise to make architectural decisions (better to discuss major decisions with others).
AI-generated code must be reviewed for maintainability, efficiency, and alignment with coding standards.
You are fully responsible for the final code output, even when using LLMs.
Never input sensitive information into AI tools. This includes API keys, passwords, customer data, proprietary algorithms, etc.

exploring different approaches and learning complex concepts.
improving productivity (e.g., code completion, refactoring, bug detection).
generating comments and docstrings, but not general documentation. AI generates too much meaningless words and you will be punished for creating such documentation.

submitting LLM’s output without understanding, review and modification.
delegating architectural decisions, complex business logic, or entire feature development without oversight.
sharing data that should never be shared anywhere in the internet.
using AI without understanding of programming, frameworks and coding best practices.